Insights Into the Genetic Basis of Predator-Induced Response in Daphnia Galeata

bioRxiv preprint doi: https://doi.org/10.1101/503904; this version posted June 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

Insights into the genetic basis of predator-induced response in Daphnia galeata

Verena Tams*1, Jana Helene Nickel*2, Anne Ehring2, Mathilde Cordellier2

* equal contribution 1 Universität Hamburg, Institute of Marine Ecosystem and Fishery Science, Große Elbstraße 133, 22767 Hamburg, Germany 2 Universität Hamburg, Institute of Zoology, Martin-Luther-King-Platz 3, 20146 Hamburg, Germany

Corresponding author: [email protected], phone +49 (0)40-42838-3933 Abstract

Phenotypic plastic responses allow organisms to rapidly adjust when facing environmental challenges - these responses comprise morphological, behavioral but also life-history changes. Alteration of life-history traits when exposed to predation risk have been reported often in the ecological and genomic model organism Daphnia. However, the molecular basis of this response is not well understood, especially in the context of fish predation. Here, we characterized the transcriptional profiles of two Daphnia galeata clonal lines with opposed life histories when exposed to fish kairomones. First, we conducted a differential gene expression, identifying a total of 125 candidate transcripts involved in the predator-induced response, uncovering substantial intra-specific variation. Second, we applied a gene co- expression network analysis to find clusters of tightly linked transcripts revealing the functional relations of transcripts underlying the predator-induced response. Our results showed that transcripts involved in remodeling of the cuticle, growth and digestion correlated with the response to environmental change in D. galeata. Furthermore, we used an orthology- based approach to gain functional information for transcripts lacking gene ontology (GO) information, as well as insights into the evolutionary conservation of transcripts. We could show that our candidate transcripts have orthologs in other Daphnia species but almost none in other arthropods. The unique combination of methods allowed us to identify candidate transcripts, their putative functions and evolutionary history associated with predator- induced responses in Daphnia. Our study opens up to the question as to whether the same molecular signature is associated fish kairomones-mediated life-history changes in other Daphnia species. Keywords: RNA-seq, predator-induced response, gene co-expression, phenotypic plasticity, Daphnia galeata

1 bioRxiv preprint doi: https://doi.org/10.1101/503904; this version posted June 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

widely used in ecology, evolution and Introduction ecotoxicology (e.g. Lampert 2011; Miner et Organisms are challenged al. 2012; Picado et al. 2007). Members of throughout their lives by environmental this genus link trophic levels from primary changes that have an impact on the health producers to consumers in freshwater and fitness of each individual. A given ecosystems and are, therefore, vulnerable phenotype that is advantageous in one to high predation risk (Lampert 2011). environmental setup might become Extensive shifts in behavior, morphology disadvantageous in another. In general, and life history traits have been observed organisms have two possibilities to cope in response to predation and predation with environmental changes: return to the risk. The responses induced by ecological niche by behavioral (i.e. invertebrate predators include migration) or physiological changes, or morphological changes such as the change the boundaries of their ecological formation of helmets in D. cucullata niche by genetic adaptation (Van Straalen (Agrawal et al. 1999) and D. longispina 2003.). The former is achieved at the (Brett 1992) and the formation of neck phenotypic level and described as a teeth in D. pulex (Tollrian 1995). Vertebrate phenotypic plastic response, while the predators cues have been shown to induce latter is a genetic adaptation process, behavioral changes linked to diel vertical where genotypes with a higher fitness pass migration (Cousyn et al. 2001; Effertz & von on their alleles to the next generation. Elert 2017; Hahn et al. 2019) as well as Predation is an important biotic changes in life history traits (Boersma et al. factor structuring whole communities (e.g. 1998; Effertz & von Elert 2017) in D. Aldana et al. 2016; Boaden & Kingsford magna. The specificity of such predator- 2015), maintaining species diversity (e.g. induced responses by vertebrate and Estes et al. 2011; Fine 2015) and driving invertebrate kairomones has been shown, natural selection in populations (e.g. e.g. for the D. longispina species complex Kuchta & Svensson 2014; Morgans & Ord from the Swiss lake Greifensee (Wolinska 2013). Vertebrate as well as invertebrate et al. 2007). The documented changes in aquatic predators release kairomones into life history traits included a decrease in size the surrounding water (Macháček 1991; at maturity when exposed to fish Schoeppner & Relyea 2009; Stibor 1992; kairomones and an increase when exposed Stibor & Lüning 1994). In some instances, to kairomones of the phantom midge kairomones can be detected by their prey, larvae, a predatory invertebrate of the inducing highly variable as well as genus Chaoborus. predator-specific responses to reduce their Although phenotypic plastic vulnerability. These predator-induced responses to predation risk have been responses are a textbook example of extensively studied in the ecological and phenotypic plasticity (Tollrian & Harvell genomic model organism Daphnia, their 1999) and have been reported in detail for genetic basis is not well understood a variety of Daphnia species (e.g. Boeing et (Mitchell et al. 2017). Linking predator- al. 2006; Herzog et al. 2016; Weider & induced responses to the underlying Pijanowska 1993; Yin et al. 2011). genome-wide expression patterns has Daphnia are small branchiopod been attempted from different crustaceans and are a model organism perspectives (length of exposure time,

2 bioRxiv preprint doi: https://doi.org/10.1101/503904; this version posted June 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

species and experimental conditions) in network analysis lies in the opportunity to Daphnia. Orsini et al. (2018) investigated correlate gene expression and external the effect of short-term exposure to fish information (Langfelder & Horvath 2008) kairomones (several hours) in D. magna, thus simplifying the process of candidate revealing no change in gene expression. genes identification. Additionally, Yet another study identified over 200 orthology analysis allows revealing differentially expressed genes in response functional roles as well as the evolutionary to invertebrate predation risk in D. pulex, history of transcripts. The degree of of which the most prominent classes of conservation of the predator-induced upregulated genes included cuticle genes, response can be estimated by finding zinc-metalloproteinases and vitellogenin orthologous genes in species having genes (Rozenberg et al. 2015). Finally, a diverged million years ago (Cornetti et al. study on D. ambigua under vertebrate 2019). We hypothesize that a common predation risk revealed ~50 responsive predator-induced response exists within genes involved in reproduction, digestion one Daphnia species at the gene and exoskeleton structure (Hales et al. expression level and that transcripts 2017). involved are evolutionary conserved among Daphnia species under vertebrate Our goal is to investigate the predation risk. genetic basis of life history shifts in response to vertebrate predation risk. Materials and methods Daphnia galeata is the ideal candidate Experimental organisms species is to address this question, since this species does not show diel vertical This study was conducted on two D. migration behavior (Stich & Lampert 1981) galeata genotypes originally hatched from or severe morphological changes in the resting eggs collected from Müggelsee presence of vertebrate predator cues even (northeast Germany). A previous study after long exposure, but diverse shifts in involving 24 genotypes (clonal lines) from life history traits under vertebrate four different lakes revealed that the predation risk (Tams et al. 2018). With a variation for some life history traits combined approach, we aim to understand increased for genotypes exposed to fish the complexity of responses to kairomones within the Müggelsee environmental changes such as those population (Tams et al. 2018), meaning a induced by predators, which are known to broader range of phenotypes were vary across Daphnia species. We applied a displayed for that life history trait. We transcriptomic approach (RNA- chose the genotypes M6 and M9 which sequencing), followed by differential gene differed in all of their life history traits and expression, gene co-expression and were at the contrasting ends of the orthology analysis. Gene co-expression phenotypic range exhibited by D. galeata network analysis allows to infer gene exposed to fish kairomones. Genotype M6 functions because of the modular structure displayed a phenotype which matured of co-expressed genes and their functional later, produced less offspring and stayed relations; often co-expressed genes share smaller under predation risk. Genotype M9 conserved biological functions (Bergmann displayed the opposite phenotype, i.e. et al. 2003; Subramanian et al. 2005). A matured earlier, produced more offspring further benefit of the co-expression

3 bioRxiv preprint doi: https://doi.org/10.1101/503904; this version posted June 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

and became larger under predation risk kairomone concentration. The algae (Figure 1). concentration was calculated from the photometric measurement of the Media preparation absorbance rate at 800 nm. Cetyl alcohol ADaM (Klüttgen et al. 1994) was was used to break the surface tension used as the basic medium for fish and during breeding and the experiment to Daphnia cultures. Two types of media, fish reduce juvenile mortality (Desmarais kairomone and control, were prepared and 1997). Breeding and experimental phases used for breeding and experimental were conducted at a temperature of 20°C conditions as detailed in Tams et al. (2018). and a 16h light / 8h dark cycle in a brood Briefly, fish kairomone medium was chamber with 30% of its maximum light obtained by maintaining five ide (Leuciscus intensity (Rumed, Type 3201D). idus) in a 20L tank for 24 hours prior to Experimental design and procedures medium use. All media were filtered (Whatman, membrane filters, ME28, Each genotype was bred in Mixed cellulose-ester, 1.2µm) prior to use kairomone-free medium (control) and in and supplemented with 1.0 mg C L-1, P rich fish kairomone medium (predation risk) for Acutodesmus obliquus. Media was two subsequent generations before the exchanged daily (1:2) to ensure a nutrient- start of the experiment to minimize inter- rich environment and a constant fish individual variances. To this end, 20 egg-

Figure 1: Reaction norms of selected life history traits of experimental genotypes (mean +/- SE) from a previous experiment (Tams et al 2018) emphasizing the opposing environmental effect on life history traits for the two genotypes. (a) Age at first reproduction in days. (b) Total number of broods per female (b) total number of offspring per female (d) somatic growth rate in µm per day. ‘Blue’: genotype M6. ‘Orange’: genotype M9. ‘Control’: environment without fish kairomone exposure. ‘Fish’: environment with fish kairomone exposure (predation risk).

4 bioRxiv preprint doi: https://doi.org/10.1101/503904; this version posted June 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

bearing females per genotype were Appropriate amounts of RNA were randomly selected from mass cultures. not available from single individuals, and From these females of unknown age, hence we used pools of experimental neonates (<24h) were collected and raised individuals. Similar pooling approaches under experimental conditions in 750 mL have been used in other Daphnia beakers at densities of <40 neonates per differential gene expression studies (Hales beaker. They served as grandmothers (F0) et al. 2017; Herrmann et al. 2018; for the experimental animals (F2). Based Huylmans et al. 2016; Orsini et al. 2018; upon previous work (Tams et al. 2018), we Ravindran et al. 2019; Rozenberg et al. started the second (F1) generation after 2015). Only experimental females bearing 16-20 days to ensure that offspring from eggs were pooled, resulting in a minor the 3rd to 5th brood were used to start the difference in age and experimental time as next generation. The third generation of some experimental females had been experimental individuals (F2) was started pooled a day later. The advantage of after 18 days. At the start of the sampling females in their inter-molt stage experiment, a pair of neonates was (egg-bearing) is to ensure a stable gene introduced in the experimental vessels (50 expression (Altshuler et al. 2015). Total mL glass tube) to compensate for juvenile RNA was extracted from pools of 20 egg- mortality. Before the release of the first bearing adults after homogenizing in brood, on day 6, one of the individuals was RNAmagic (Bio-Budget technologies, randomly discarded whenever necessary Krefeld, Germany) for 5 min with a so that only one individual remained in disposable pestle and a battery-operated each vessel. During the 14 days of the homogenizer. Samples were stored at – experiment, neonates were removed every 80°C until RNA isolation. Chloroform was 24 hours and the number of broods of each added to the homogenate before experimental female was documented centrifuging in Phasemaker tubes (Thermo before media renewal. The adult females Fisher Scientific, Carlsbad, CA, USA) to (F2) were pooled (n=20) and homogenized separate the upper aqueous and lower in RNAmagic (Bio-Budget technologies, phenol phase. The upper aqueous phase Krefeld, Germany). Five biological was transferred into a clean replicates were produced per experimental microcentrifuge tube and the RNA condition (environment) and per genotype, precipitated with absolute ethanol. RNA resulting in a total of 400 individuals (two purification and DNAse treatment were genotypes x two environments x 20 conducted following the Direct-zolTM RNA individuals x five biological replicates). Two MiniPrep Kit protocol (Zymo Research, of these replicates were backup in case of Irvine, CA, USA) with slight modifications. a downstream failure, and three were Quality and quantity of purified RNA was processed for sequencing (see below). The checked by spectrophotometry using a experiment lasted for 14 days for each NanoDrop 2000 (Thermo Fisher Scientific, experimental individual to assess the long- Wilmington, DE, USA). The RNA integrity term effect of fish kairomones on gene was confirmed with the Agilent expression level in D. galeata. TapeStation 4200 (Agilent Technologies, Santa Clara, CA, USA). Only samples Data collection and analysis showing no degradation and RNA Integrity RNA isolation and preparation Numbers (RIN) > 7 were used for

5 bioRxiv preprint doi: https://doi.org/10.1101/503904; this version posted June 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

subsequent steps. Sequencing was v.0.9.1 was used to quantify the number of performed for 12 samples (two genotypes reads mapped to each transcript (Anders et x two environments x three biological al. 2015). replicates). Differential gene expression analysis RNA-seq library construction and Differential gene expression sequencing analysis was performed in the R Library construction and environment v.3.4.2 (R Core Team 2017) sequencing was identical for all samples with the R package ‘DESeq2’ v.1.18.1 (Love and was performed by the company et al. 2014) implemented in Bioconductor Macrogen (Seoul, South Korea). RNA-seq v.3.6 (Gentleman et al. 2004). The libraries were constructed using Illumina calculation was based on normalized read TruSeq library kits. Illumina HiSeq4000 (San counts per environment (control & fish) Diego, CA, USA) instrument was used for using negative binomial generalized linear paired-end sequencing with 101-bp read models. Prior to the analysis, all transcripts length resulting in 48-79 million reads per with a read count lower than 12 across all library. libraries were excluded. Results were filtered post-hoc by an adjusted p-value RNA-seq quality control and mapping (padj < 0.05) (Benjamini & Hochberg 1995) The quality of raw reads was to reduce the false discovery rate (FDR) and checked using FastQC v.0.11.5 (Andrews filtered for a log2 fold change ≥ 1. 2010). Adapter trimming and quality Differentially expressed transcripts (DETs) filtering were performed using were binned into four groups: <2-fold, 2- to Trimmomatic v.0.36 (Bolger et al. 2014) 4-fold, 4- to 6-fold and >6-fold difference in with the following parameters: expression. The three biological replicates ILLUMINACLIP: TruSeq3-PE.fa:2:30:10 were checked for homogeneity by principal TRAILING: 20 SLIDINGWINDOW: 4:15. After component analysis (PCA). A differential trimming, the read quality was checked expression analysis of genes between again with FastQC to control for the environments, between genotypes and successful removal of adapters. The between environments within each cleaned reads were mapped to the genotype was done. In addition, a two- reference transcriptome of D. galeata factor analysis was applied to investigate (Huylmans et al. 2016) using NextGenMap genotype-environment interactions (GxE). v.0.5.4 (Sedlazeck et al. 2013) with PCA plots were created in R with ‘ggplot2’ increased sensitivity (--kmer-skip 0 –s 0.0). v.2.2.1 (Wickham 2009). The web tool All reads which had an identity < 0.8 and jvenn (Bardou et al. 2014) was used to mapped with a residue number < 25 were visualize the number of shared transcripts reported as unmapped. The option ‘strata’ between groups. was used to output only the highest Gene co-expression network analysis mapping scores for any given read and thus the uniquely mapped reads. The quality of Variance-stabilized read counts filtering and mapping reads was verified obtained from the previous ‘DESeq2’- with QualiMap v.2.2.1 (Okonechnikov et al. analysis were used in the co-expression 2016). Subsequently, the htseq-count analysis. First, an automatic, signed python script implemented in HTSeq weighted, single gene co-expression

6 bioRxiv preprint doi: https://doi.org/10.1101/503904; this version posted June 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

network was constructed via an adjacency OrthoMCL cluster information from matrix in the R environment v.3.2.3 with the reference transcriptome of D. galeata the R package ‘WGCNA’ v.1.61 (Langfelder (Huylmans et al. 2016) was used to & Horvath 2008). Second, gene co- enhance our understanding of the expression modules – cluster of highly functional roles of the transcripts of interconnected genes – were identified interest and their evolutionary history. based on the topological overlap matrices ‘OrthoMCL’ is a tool to identify clusters of (TOM) with a soft cut-off threshold of 14 in homologous sequences in multiple species ‘WGCNA’. Module eigengenes (ME) – i.e. orthologs. When assuming that representing the average gene expression orthologs are functionally conserved, of their module – were calculated and used known functions of orthologs in one to investigate their relationship with other species can be used to assign putative modules as well as external information functions to sequences from other species (predation risk and genotype). ME-trait in the same orthologous group (Li et al. correlations were calculated to identify 2003). These OrthoMCL clusters were transcripts of interest with correlation originally build based on data for three values of > 0.5 or < -0.5. Finally, hubgenes Daphnia species (D. galeata, D. pulex and – defined as the most interconnected D. magna) and two insect species genes per module – were identified to gain (Drosophila melanogaster and Nasonia insight into the biological role of a gene co- vitripennis). Further details about the expression module. genome versions and annotations are available in the original publication by Gene set enrichment analysis (GSEA) (Huylmans et al. 2016). To identify potential function of We analyzed the OrthoMCL clusters differentially expressed and co-expressed containing transcripts of interests by transcripts, we assigned Gene Ontology counting how many orthologs from other (GO) annotations using the reference species where comprised in these clusters. transcriptome of D. galeata (Huylmans et Clusters were grouped into different al. 2016). To shed light on the biological categories: Daphnia galeata specific, importance of transcripts of interest, we Daphnia galeata plus one of the other performed a gene set enrichment analysis Daphnia species (D magna or D. pulex), in R with the package ’topGO’ v.2.30.0 Daphnia specific, and “all-species” for (Alexa & Rahnenfuhrer 2016). The default those containing at least one transcript for algorithm 'weight01' was used taking the each of the reference species (three hierarchy of GO terms into account, which Daphnia and two insect species). This results in fewer false positive results (Alexa allowed to measure how conserved our & Rahnenfuhrer 2016). Given that, a signal was and whether the response to multiple testing correction after the predator-risk was affecting transcripts Fisher's exact test was not applied specific to this particular species. (Timmermans et al.). GO terms of the three GO categories 'Molecular Function' (MF), Results 'Biological Process' (BP) and ‘Cellular RNA-seq data quality Compounds’ (CC) with a p-value < 0.05 were considered significant. RNA samples passed all quality steps before RNA sequencing. All 12 Orthology analysis samples were successfully sequenced,

7 bioRxiv preprint doi: https://doi.org/10.1101/503904; this version posted June 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

resulting in 48.2 to 79.2 million reads of mapped to the D. galeata reference 101 bp length. After trimming and quality transcriptome. After the filtering process, control ~90% of trimmed reads were kept the full dataset used for further analysis for further analysis. An average of 88% of comprised a total of 32,903 transcripts. these trimmed reads were uniquely

Table 1. Number of differentially expressed transcripts (DETs) in D. galeata (p.adj=0.05, foldchange= log2). (A) Results of the one-factor analysis. ‘Genotype’: DETs between genotypes (M6 over M9). ‘M6’: DETs within genotype M6 between environments (fish over control). ‘M9’: DETs within genotype M9 between environments (fish over control). (B) Results of the two-factor analysis. ‘M6’: environment effect for genotype M6 (fish over control). ‘M9’: environment effect for genotype M9 (fish over control). ‘M6 vs M9’: differences between the two genotypes in control environment (M6 over M9). ‘M6 vs M9 FK’: differences between genotypes in fish environment (M6 over M9). ‘GxE’: genotype-environment interaction (genotype x predation risk).

A All <2-fold 2- to 4-fold 4- to 6-fold > 6-fold Genotype 5283 1964 1486 927 906 up 2228 743 630 410 445 down 3055 1221 856 517 461 M6 30 11 11 6 2 up 3 3 0 0 0 down 27 8 11 6 2 M9 57 24 27 5 1 up 21 16 5 0 0 down 36 8 22 5 1

B All <2-fold 2- to 4-fold 4- to 6-fold > 6-fold M6 4 1 2 0 1 up 1 0 0 0 1 down 3 1 2 0 0 M9 68 45 16 6 6 up 29 22 5 1 1 down 39 23 11 5 0 M6 vs M9 4687 1624 1204 899 960 up 1990 633 494 405 458 down 2697 991 710 494 502 M6 vs M9 FK 3820 1114 915 826 965 up 2016 611 478 428 499 down 1804 503 437 398 466 GxE 22 11 6 4 1 up 7 3 4 0 0 down 15 8 2 4 1

8 bioRxiv preprint doi: https://doi.org/10.1101/503904; this version posted June 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

Differential gene expression analysis grouping of read counts and to help identify batch effects. The first principal Before subsequent analysis, all component (PC 1) explained 83% of the transcripts with a read count lower than 12 variance between genotypes, revealing no across all libraries were excluded. 23,982 clear clustering of read counts per transcripts remained for both genotypes: environment (Figure 2A). PC 2 explained 21,740 transcripts for genotype M6 and just 10% of the variance, which seems to be 21,813 for genotype M9. related to the variance between replicates. A principal component analysis Separate plots per genotype improved the (PCA) was performed to visualize the visualization of replicate and

Figure 2: (A) Principal component (PC) plot of the biological RNA-seq samples in D. galeata. Yellow: control environment. Blue: fish environment (predation risk). Triangles: genotype M9. Circles: genotype M6. (B) Venn diagram of the 125 differentially expressed transcripts (DETs) related to predation risk in D. galeata. The set of DETs originates from the one and two factor analysis. ‘M6 (one)’: DETs from the one-factor analysis for the genotype M6. ‘M9 (one)’: DETs from the one-factor analysis for the genotype M9. ‘M6 (two)’: DETs from the two-factor analysis for the genotype M6. ‘M9 (two)’: DETs from the two-factor analysis for the genotype M9.

9 bioRxiv preprint doi: https://doi.org/10.1101/503904; this version posted June 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

environmental differences (Appendix 1) gene co-expression modules (Figure 4, but did not indicate an evident clustering Table 2). A total of eight modules by environment either. correlated to predation risk or genotype with a correlation coefficient >0.5 or < -0.5 The differential expression analysis (Appendix 3). Three small gene co- revealed no differentially expressed expression modules associated with transcripts (DETs) between environmental predation risk: ‘salmon’ (n= 107), ‘red’ (n= groups, but a total of 5,283 DETs between 519) and ‘tan’ (n= 116). The ‘salmon’ genotypes (up: 2,228 (42%), down: 3,055 module correlated positively with (58%); Figure 3). Because of the strong predation risk (p= 0.01); the ‘red’ and the genotype effect, the genotypes were ‘tan’ module correlated negatively with analyzed separately in a one-factor analysis predation risk (pred= 0.03, ptan= 0.05). Five (Table 1A). For genotype M6, there were gene co-expression modules associated to 30 DETs between environments (up: 3 genotype. The two large co-expression (10%), down: 27 (90%)). For genotype M9, modules ‘turquoise’ (n= 5,154) and ‘brown’ there were 57 DETs between environments (n= 4,760) correlated positively with (up: 21 (37%), down: 36 (63%)). A two- genotype (pturquoise= 0.05, pbrown> 0.001), factor analysis accounted for the genotype- while ‘blue’ (n= 4,868), ‘yellow’ (n= 4,612) environment interaction (GxE) (Table 1B). and ‘green’ (n= 950) correlated negatively Between environments genotype M6 had with genotype (pblue< 0.001, pyellow= 0.04, four DETs (up: 1 (25%), down: 3 (75%)) and pgreen= 0.04). A dendrogram of the genotype M9 had 68 DETs (up: 29 (43%), relationship of all co-expression modules down: 39 (57%)). The GxE resulted in 22 showed that the co-expression modules DETs (up: 7 (32%), down: 15 (68%)). ‘blue’, ‘green’ and ‘yellow’ related closely No DETs were shared between the to genotype and ‘salmon’ to predation risk two genotypes under predation risk; DETs (Figure 5). were shared between the one and two The most highly interconnected factor analysis only within one genotype gene within a gene co-expression module (Figure 2B). In total 125 transcripts were (hubgene) was identified for each module differentially expressed between the two (Table 2). Three hubgenes of co-expression environments (hereafter, predation risk- modules belonged to the previously related DETs) (up: 40 (32%), down: 85 identified predation risk-related DETs, (68%); Figure 3, Appendix 2). The namely those for the co-expression differential expression was strong (fold modules ‘midnightblue’, ‘salmon’ and ‘tan’ change >2) for downregulated DETs (~60% (Table 2). for predation risk and genotype-related DETs) and for ~67% of upregulated the In total 104 of 125 predation risk- genotype-related DETs (Table 1, Figure 3). related transcripts identified through the Only ~28% of upregulated, predation risk- differential expression analysis belonged to related DETs were strongly differentially a co-expression modules of interest expressed. (‘salmon’ n=13, ‘tan’ n=9, ‘red’ n=3, ‘turquoise’ n=21, ‘brown’ n=17, ‘blue’ n=7, Gene co-expression network analysis ‘green’ n=1, ‘yellow’ n=33; Appendix 2). The single network analysis clustered the expressed transcripts into 16

10 bioRxiv preprint doi: https://doi.org/10.1101/503904; this version posted June 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

‘brown’, ‘green’ and ‘yellow’ (total n= 20,344). 36% of transcripts deriving from the co-expression modules of interest were annotated (‘turquoise-blue-brown-greenyellow’ n= 7,117; ‘tan-red-salmon’ n= 207). The lowest rate of annotation (23%) was for genotype-related DETs (n= 1,230 of 5,284) and the highest (33%) for the predation risk-related DETs (n= 41 of 125). Five out of the 15 hubgenes had a GO annotation (Table 2). Gene set enrichment analysis (GSEA) A gene set enrichment analysis was performed on either predation risk-related transcripts of interest (predation risk- related DETs plus transcripts of predation risk-related co-expression modules (‘salmon’, ‘tan’, ‘red’)) or genotype-related (genotype-related DETs plus transcripts of genotype-related co-expression modules (‘turquoise’, ‘blue’, ‘brown’, ‘green’, ‘yellow’)). In total 44 GO terms were significantly enriched in the predation risk- Figure 3: Up- and downregulated differentially related transcript set; 29 of these GO terms expressed transcripts (DETs) grouped by expression were unique (Appendix 4). In the genotype- foldchange. (A) DETs related to genotype. (B) DETs related transcript set 209 GO terms were related to predation risk. significantly enriched; 168 of them were Gene ontology (GO) annotation unique (Appendix 5). There were only eight unique significantly enriched GO terms In total, 10,431 transcripts in the D. shared between the predation risk- and galeata reference transcriptome had Gene genotype-related transcript sets of Ontology (GO) annotations (Huylmans et interest: ‘serine-type endopeptidase al. 2016). The transcript sets of interest activity’ (GO:0004252), ‘extracellular were either predation risk- or genotype- matrix structural constituent’ related. Predation risk-related transcripts (GO:0005201), ‘cysteine-type peptidase of interest originated from the co- activity’ (GO:0008234), ‘structural expression modules ‘salmon’, ‘tan’ and constituent of cuticle’ (GO:0042302), ‘red’ (total n= 742), and the differential ‘proteolysis’ (GO:0006508), ‘homophilic gene expression analysis (one and two cell adhesion via plasma membrane factor analysis; total n= 125). Genotype- adhesion molecules’ (GO:0007156), related transcripts originated from the co- ‘oxidation-reduction process’ expression modules ‘turquoise’, ‘blue’, (GO:0055114) and ‘collagen trimer‘ (GO:0005581).

Table 2. Overview of gene co-expression modules in D. galeata. The table summarizes module color, total number of transcripts per module, the name of the most inter-connected transcript (hubgene), gene significances (GS) and its p-value for predation risk (fish kairomone exposure) and genotype as well as differentially expressed transcripts (DETs) and gene ontology (GO) IDs and classes. The module 'grey' contains all co-expressed genes which were not assigned to a co-expression module (n=1,525 (6%)). Significant p-values (p<0.05) are highlighted in bold.

GS. total number hub-gene of co-expression predation p.GS GS. p.GS module of transcripts module risk predation risk genotype genotype DETs GO.ID GO.class turquoise 5154 Dgal_a239 0.44 0.15 -0.51 0.09 no blue 4868 Dgal_sd37687381411 -0.02 0.95 1.0 0.00 no brown 4760 Dgal_s384083 0.00 0.99 -1.0 0.00 no yellow 4612 Dgal_o2422d15233t1 -0.37 0.23 0.58 0.05 no GO:0005515 protein binding green 950 Dgal_o7683t4 0.06 0.86 0.57 0.05 no GO:0042302 structural constituent of cuticle red 519 Dgal_o2656t3 -0.54 0.07 -0.14 0.67 no GO:0055114 oxidation-reduction process no GO:0004497 monooxygenase activity no GO:0005507 copper ion binding no GO:0016715 oxidoreductase activity,….. black 491 Dgal_o12661t5 -0.43 0.17 -0.12 0.64 no GO:0005515 protein binding pink 251 Dgal_t25528c0t5 0.19 0.55 0.24 0.46 no magenta 198 Dgal_t24643c0t2 0.38 0.22 0.41 0.19 no purple 181 Dgal_o698d42270t2 0.02 0.95 0.50 0.10 no greenyellow 127 Dgal_o21585d23838t2 0.04 0.89 0.61 0.04 no GO:0005509 calcium ion binding no GO:0004623 phospholipase A2 activity no GO:0016042 lipid catabolic process tan 116 Dgal_t6156c0t1 -0.55 0.06 0.13 0.69 yes salmon 107 Dgal_a 84_f_262622 0.65 0.02 0.03 0.93 yes cyan 67 Dgal_t32639c0t1 -0.15 0.64 0.29 0.36 no midnightblue 56 Dgal_a 80_j_452081 -0.43 0.16 -0.04 0.91 yes Genes not assigned to a grey 1525 module no

12 bioRxiv preprint doi: https://doi.org/10.1101/503904; this version posted June 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

Orthology analysis interest were distributed among 563 orthoMCL clusters (2,131 D. galeata GO term and orthoMCL cluster transcripts; GO annotation n= 224). Most information was available for a total of of these orthoMCL clusters comprise 9,172 D. galeata transcripts. Out of the 867 orthologs for all three Daphnia species transcripts in the predation risk-related set (Figure 6), hinting at a common Daphnia 600 were assigned to an orthology cluster. Predation risk-related transcripts of

Figure 4: Cluster dendrogram of transcripts in Daphnia galeata, with dissimilarity based on the topological overlap matrices (TOM). Additional assignments are module colours, the gene significances (GS) for the trait genotype and predation risk (fish kairomone exposure). ‘Red’: positive correlation of the module with the respective trait. ‘Blue’: negative correlation of the module with the respective trait. Darker hues indicate higher correlation values (darkest hue = 1; lowest hue = 0).

13 bioRxiv preprint doi: https://doi.org/10.1101/503904; this version posted June 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

response. basis of predator-induced responses, we performed gene expression profiling on two D. galeata genotypes after long-term exposure to fish kairomones simulating predation risk. We identified a number of transcripts correlating to predation risk and used gene co-expression network analysis, gene ontology annotation and gene set enrichment analysis to describe their putative biological functions. The orthology analysis provided insights into the evolutionary conservation of transcripts, indicating that the majority of transcripts involved in predation risk response were Daphnia specific. Insights from differential gene expression Figure 5: Cluster dendrogram of module analysis – similar transcripts, differentially eigengenes and external traits genotype and expressed? predation risk. This dendrogram shows the relationship between the co-expression In contrast to our expectations, the modules which are here represented by its differential gene expression analysis module eigengene. In addition, external revealed only a moderate number of information genotype and predation risk were included to show their relationship with the differentially expressed transcripts (DETs) gene co-expression modules. between environments within each genotype, but a large divergence between Discussion genotypes of the same population in D. From an ecological point of view, galeata. Genotype-specific molecular predator-induced responses in Daphnia responses to environmental cues were have been studied extensively. In the past reported for genotypes originating from years, few studies addressed the link different populations for D. pulex (De between such ecological traits and the Coninck et al. 2014) as well as for D. magna underlying genetic pathways (Hales et al. (Orsini et al. 2018). Since we explicitly 2017; Orsini et al. 2018; Rozenberg et al. chose genotypes from one population to 2015). Similar trends in life history shifts minimize the potential genetic variation, after exposure to predator kairomones population origin can be excluded as an have been observed across Daphnia explanation for the observed intraspecific species showing e.g. the predominant divergence in gene expression profiles in D. trend of early maturation and a decreased galeata, concurring with previous studies body size under vertebrate predation risk in our group (Ravindran et al. 2019). (e.g. Riessen 1999). Thereupon, it seems Instead, the apparent genotype-specific reasonable to formulate the hypothesis response in our study might be explained that similar transcripts, differentially by the phenotypic divergence between the expressed, could be involved in the two studied clones. predator-induced response in all Daphnia One explanation for the low number of species. To gain insights into the genetic DETs concerning predation risk

14 bioRxiv preprint doi: https://doi.org/10.1101/503904; this version posted June 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

(environment) compared to genotype- genotypes display transgenerational specific differences is that life history plasticity and/or pass on epigenetic changes could only marginally correlate modifications after exposure to fish with gene expression. The D. galeata kairomones. Further investigations are genotypes used here only displayed shifts therefore required to understand the in life history, whereas other Daphnia epigenetic level of inheritance in Daphnia. species show additional adaptations of We expected to find similar morphology and behavior that could be transcripts to be involved in the contrasting caused by or correlated to much stronger life history responses of the two genotypes differential gene expression, e.g. neck- under predation risk. In contrast, a teeth induction that was linked to 230 completely different set of transcripts was differentially expressed genes in D. pulex linked to predation risk within each (Rozenberg et al. 2015).The moderate genotype. The most likely explanation is number of DETs found under predation risk the high variation in the biological could be explained by other causes than replicates which resulted in no clear gene expression as well; additional distinction between environments. To posttranslational processes, such as clarify whether DETs are actually genotype- miRNA-mediated regulation or increased specific it would be necessary to generate degradation (Schwarzenberger et al. 2009) RNA-seq data for more D. galeata might play a role. genotypes from the same and other There were three reasons why we populations, both with shared and expected more pronounced and divergent life histories. cumulative changes in differential gene Insights from gene co-expression and gene expression in the third experimental set enrichment analysis (GSEA) – different D. galeata generation. First, the chosen transcripts, different functions? genotypes displayed strong shifts in life history traits after three generations of fish Although different transcripts were kairomone exposure (Tams et al. 2018). identified in the differential expression Second, the effect of kairomone exposure analysis, the combined approach revealed is expected to be cumulative and to their similarity of biological functions. In increase over the course of multiple brief, our gene co-expression and gene set generations; e.g. D. pulex displays the enrichment analysis revealed digestion- largest helmets when exposed to and growth-related enriched GO terms. kairomones from Leptodora kindtii (an These are interesting because predator- invertebrate predator) for two generations induced responses in Daphnia include compared to the first generation (Agrawal changes in body size (e.g. Tams et al. 2018) et al. 1999). Third, transgenerational and morphological modifications (e.g. plasticity was described in D. ambigua Laforsch & Tollrian 2004). Such (Hales et al. 2017); genes were significantly modifications require energy allocated differentially expressed after one from nutrients; consequently digestive generation of fish kairomone exposure enzymes like peptidases have been shown (n=48 DEGs) and without kairomone to be important for juvenile growth rate in exposure after the second (n=223 DEGs) D. magna (Schwarzenberger et al. 2012). and third (n=170 DEGs) generation. To Transcripts of the ‘salmon’ gene co- date, it is unknown whether D. galeata expression module in D. galeata were

15 bioRxiv preprint doi: https://doi.org/10.1101/503904; this version posted June 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

significantly enriched for ‘serine-type genotypes did not display pronounced endopeptidase activity’, the most morphological defenses, but changes in important digestive protease in the gut of body size and symmetry especially with D. magna (Agrawal et al. 2005). The regard to head shape (Tams et al. 2018). exposure to predator kairomones for one Furthermore, for D. magna, D. pulex and D. generation in D. ambigua – a species from cucullata, not only visible morphology the D. pulex-complex more closely related changes but also fortification of the to D. galeata than D. magna (Adamowicz et carapace in the presence of predator al. 2009; Cornetti et al. 2019) – led to an kairomones has been recorded (Laforsch & up-regulation of genes related to digestive Tollrian 2004; Rabus et al. 2013). Our functions (Hales et al. 2017). results indicated that ultrastructural Cyanobacterial protease inhibitors cause defenses could also be present in D. considerable damage to Daphnia galeata. populations by inhibiting the gut Altogether, cuticle-associated proteases, impairing their digestion proteins seem to play an essential role in (Schwarzenberger et al. 2010). These the response to vertebrate or invertebrate studies concord with our results suggesting predators in Daphnia. DETs found in that an increase in ‘serine-type genotype M6 showed the possible endopeptidase activity’ leads to improved involvement of ‘metallocarboxypeptidase digestion and feeding efficiency that is activity’, known to be involved in the stress necessary for the resource allocation that response to copper in D. pulex (Chain et al. comes with shifts in life history, such as 2019). producing a greater number of offspring. Interestingly, ‘chitin metabolic The GO term ‘structural constituent process’, ‘proteolysis’, ‘structural of cuticle’ was significantly enriched in constituent of cuticle’, ‘chitin binding’, both genotypes suggesting that even ‘serine-type endopeptidase’ and though there was no overlap in the ‘metallopeptidase activity’ were all found affected transcripts, similar functions were to be enriched in a gene expression affected. The ‘structural constituent of analysis during the molt cycle in the marine cuticle’ was enriched in D. pulex exposed to copepod Calanus finmarchicus (Tarrant et Chaoborus kairomones (An et al. 2018; al. 2014). Since Daphnia need to shed their Rozenberg et al. 2015) and related to rigid carapace in order to grow, molting is remodeling of the cuticle. Furthermore, it directly related to changes in body size. was also enriched in the proteomic Another analysis of D. magna exposed to response of D. magna to Triops Triops cancriformis kairomones revealed cancriformis (Otte et al. 2015) and is the role of proteins related to the cuticle, thought to be related to changes in muscular system, energy metabolism and carapace morphology as well as the regulatory proteins that may be involved in formation of ultrastructural defenses of morphological carapace defenses and the cuticle (Rabus et al. 2013). Genes changes in resource allocation (Otte et al. related to body remodeling and activation 2014). In conclusion, a number of biological of cuticle proteins were enriched for D. functions hypothesized to be involved in magna exposed to vertebrate and kairomone response could be confirmed, invertebrate predator kairomones (Orsini e.g., transcripts related to body remodeling et al. 2018). The investigated D. galeata and growth.

16 bioRxiv preprint doi: https://doi.org/10.1101/503904; this version posted June 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

It is worthwhile to mention that sequences (Pearson 2013). Here, we chose some biologically interesting gene this approach to gain insight into the functions were only found with the help of evolutionary conservation of transcripts of the gene co-expression network analysis interest. The results obtained with the and would have been overlooked with only OrthoMCL approach provide support for a differential expression analysis. For the hypothesis that phenotypic plastic example, the GO term ‘growth factor predator-induced responses might be activity’ occurred in both ‘red’ and ‘tan’ evolutionary conserved in Daphnia. modules, which correlated negatively with fish kairomone exposure and comprising transcripts not identified as DETs. Nevertheless, they could be extremely important for life history changes and might be directly related to changes in somatic growth rate and body size. For a more comprehensive understanding of genetic links to phenotypic variation and their biological functions, further annotations and therefore functional tests of candidate transcripts are needed. At present, only one third of transcripts of interest were

annotated. When GO annotations Figure 6: predation risk related orthoMCL progress, a re-analysis might provide new clusters grouped according to the species of elements for understanding the genetic origin of included transcripts - dgal= D. basis of predator-induced responses in galeata; dma= D. magna; dpu = D. pulex; all Daphnia. Onward, generating gene species = all five species included in the analysis. expression data for all 24 genotypes used in Tams et al. (2018) would allow to create However, several strategies might models to predict the effect of predation have developed over time to cope with or risk for European D. galeata. The adapt to predation risk. This hypothesis prediction of reproductive success of seems likely since Daphnia have the ability Daphnia – an ecological keystone species – to rapidly adapt to local predator regimes in response to environmental disturbances (Declerck & De Meester 2003) and our or changes is useful to forecast detrimental study provides elements supporting it. effects resulting in regime shifts within the First, the differential gene expression ecosystem (Asselman et al. 2018). analysis revealed genotype-specific Insights from orthology analysis – molecular responses to predation risk for homologous sequences (common genotypes originating from the same ancestor), evolutionary conserved? population. Second, the involved transcripts have similar functions relating Using orthology – orthologs being to life history changes induced by homologous sequences that differ because predation risk, but different transcripts of speciation events – is the most popular were involved in the predator-induced strategy to derive functional similarity of response for each genotype. This concurs

17 bioRxiv preprint doi: https://doi.org/10.1101/503904; this version posted June 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

with the suggestion of niche-specific functions – relating to digestion and adaptation in D. magna due to the growth – were identified for the genotypes genotype- and condition-specific with the same population origin under transcriptional response to environmental predation risk. The transcriptional profiling changes of biotic and abiotic factors (Orsini revealed differentially expressed et al. 2018). Their gene co-expression transcripts and gene co-expression analysis revealed that genes of interest modules in connection to predator- were crustacean related, meaning that the induced responses in D. galeata. The conservation of genes did not exceed the biological functions discovered here level of crustaceans (Orsini et al. 2018). represent a valuable starting point for Further insights into evolutionary future investigations addressing the conservation of differentially and/or co- functionality of certain transcripts per se or expressed transcripts linked to phenotypic in respect to a response to environmental traits are available for modern and ancient changes. For example, by providing (resurrected) D. pulicaria exposed to detailed lists of candidate transcripts one different phosphorous regimes (Frisch et can choose specific candidates to test their al. 2020). With a different, yet similar biological functions in knock-down approach this recent study reveals the experiments. Lastly, orthology analysis importance of a holistic approach to tackle revealed that predation risk-related the question: What is the molecular basis transcripts possess orthologs in other of phenotypic responses to environmental Daphnia species, suggesting that changes? phenotypic plastic predator-induced responses are evolutionary conserved, and Gene expression analyses to warranting further investigation. uncover the molecular basis of predation- induced phenotypic changes have focused Competing Interests Statement so far on single species, and invertebrate predators. Studies conducted in similar None declared. conditions are lacking and prevent us from Data Accessibility Statement drawing conclusions about a general Raw RNA-seq reads for all 12 Daphnia response to fish predation. In the samples and the experimental set up for future, simultaneous exposure of several the analysis of DETs are available from species to kairomones, and the coupling of ArrayExpress (accession E-MTAB-6234). phenotyping and gene expression would Raw read counts, custom scripts and help to address the question of a supplementary tables will be made conserved response. available on Dryad upon publication. Conclusion Authors’ contributions In summary, the aim of this study The study was designed by VT, JHN was to characterize the genetic basis for and MC; laboratory work was carried out the predator-induced response of the by JHN, AE and VT; gene expression and freshwater grazer D. galeata. Our gene network analysis was done by JHN hypothesis that clonal lines present a and VT; VT, JHN and MC wrote the common predator-induced response by manuscript, all authors gave final approval. regulating the expression of the same transcript set could not be confirmed. However, transcripts with similar biological

18 bioRxiv preprint doi: https://doi.org/10.1101/503904; this version posted June 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

Acknowledgments Altshuler I, McLeod AM, Colbourne We thank Jonny Schulze for his help JK, Yan ND, Cristescu ME (2015) Synergistic during Daphnia breeding and the interactions of biotic and abiotic experiment. This work was supported by environmental stressors on gene the Volkswagen Foundation (Grant No. expression. Genome 58, 99-109. 86030). Animal handling and experiments An H, Do TD, Jung G, Karagozlu MZ, were in accordance with the ethical Kim C-B (2018) Comparative transcriptome standards (approved for the execution of analysis for understanding predator- experiments on vertebrates, No. 75/15). induced polyphenism in the water flea We thank Suda Parimala Ravindran for her Daphnia pulex. Int J Mol Sci 19, 2110. advice and sharing the annotation Anders S, Pyl PT, Huber W (2015) information of the Daphnia galeata HTSeq—a Python framework to work with transcriptome. Lastly, we would like to high-throughput sequencing data. thank Jennifer Lohr for her language check Bioinformatics 31, 166-169. and very useful editing, as well as two Andrews S (2010) FastQC: A quality anonymous reviewers for comments on an control tool for high throughput sequence earlier version of this manuscript. data. Asselman J, Pfrender ME, Lopez JA, References Shaw JR, De Schamphelaere K (2018) Gene Adamowicz SJ, Petrusek A, Colbourne co-expression networks drive and predict JK, Hebert PDN, Witt JDS (2009) The scale reproductive effects in Daphnia in of divergence: A phylogenetic appraisal of response to environmental disturbances. intercontinental allopatric speciation in a Environ Sci Technol 52, 317–326. passively dispersed freshwater Bardou P, Mariette J, Escudié F, zooplankton genus. Molecular Djemiel C, Klopp C (2014) jvenn: an Phylogenetics and Evolution 50, 423-436. interactive Venn diagram viewer. BMC Agrawal AA, Laforsch C, Tollrian R Bioinformatics 15, 293. (1999) Transgenerational induction of Benjamini Y, Hochberg Y (1995) defences in animals and plants. Nature Controlling the false discovery rate: a 401, 60-63. practical and powerful approach to Agrawal MK, Zitt A, Bagchi D, et al. multiple testing. Journal of the Royal (2005) Characterization of proteases in statistical society: series B guts of Daphnia magna and their inhibition (Methodological) 57, 289-300. by Microcystis aeruginosa PCC 7806. Bergmann S, Ihmels J, Barkai N (2003) Environmental Toxicology: An International Similarities and differences in genome- Journal 20, 314-322. wide expression data of six organisms. Plos Aldana M, Maturana D, Pulgar J, Biology 2, e9. García-Huidobro MR (2016) Predation and Boaden A, Kingsford MJ (2015) anthropogenic impact on community Predators drive community structure in structure of boulder beaches. Scientia coral reef fish assemblages. Ecosphere 6, 1- Marina 80, 543-551. 33. Alexa A, Rahnenfuhrer J (2016) Boeing WJ, Ramcharan CW, Riessen topGO: Enrichment analysis for gene HP (2006) Multiple predator defence ontology. strategies in Daphnia pulex and their

19 bioRxiv preprint doi: https://doi.org/10.1101/503904; this version posted June 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

relation to native habitat. Journal of Desmarais KH (1997) Keeping Plankton Research 28, 571-584. Daphnia out of the surface film with cetyl Boersma M, Spaak P, De Meester L alcohol. Journal of Plankton Research 19, (1998) Predator-mediated plasticity in 149-154. morphology, life history, and behavior of Effertz C, von Elert E (2017) Coupling Daphnia: the uncoupling of responses. The of anti-predator defences in Daphnia: the American naturalist 152, 237-248. importance of light. Hydrobiologia 798, 5- Bolger AM, Lohse M, Usadel B (2014) 13. Trimmomatic: a flexible trimmer for Estes JA, Terborgh J, Brashares JS, et Illumina sequence data. Bioinformatics 30, al. (2011) Trophic Downgrading of Planet 2114-2120. Earth. Science 333, 301-306. Brett MT (1992) Chaoborus and fish- Fine PV (2015) Ecological and mediated influences on Daphnia longispina evolutionary drivers of geographic population structure, dynamics and life variation in species diversity. Annual history strategies. Oecologia 89, 69-77. Review of Ecology, Evolution, and Chain FJ, Finlayson S, Crease T, Systematics 46, 369-392. Cristescu M (2019) Variation in Frisch D, Becker D, Wojewodzic MW transcriptional responses to copper (2020) Dissecting the Transcriptomic Basis exposure across Daphnia pulex lineages. of Phenotypic Evolution in an Aquatic Aquat Toxicol 210, 85-97. Keystone Grazer. Molecular Biology and Cornetti L, Fields PD, Van Damme K, Evolution 37, 475-487. Ebert D (2019) A fossil-calibrated Gentleman RC, Carey VJ, Bates DM, phylogenomic analysis of Daphnia and the et al. (2004) Bioconductor: open software Daphniidae. Molecular Phylogenetics and development for computational biology Evolution 137, 250-262. and bioinformatics. Genome Biology 5, Cousyn C, De Meester L, Colbourne R80. JK, et al. (2001) Rapid, local adaptation of Hahn MA, Effertz C, Bigler L, von Elert zooplankton behavior to changes in E (2019) 5α-cyprinol sulfate, a bile salt from predation pressure in the absence of fish, induces diel vertical migration in neutral genetic changes. Proceedings of the Daphnia. Elife 8. National Academy of Sciences of the United Hales NR, Schield DR, Andrew AL, et States of America 98, 6256-6260. al. (2017) Contrasting gene expression De Coninck DI, Asselman J, Glaholt S, programs correspond with predator‐ et al. (2014) Genome-wide transcription induced phenotypic plasticity within and profiles reveal genotype-dependent across generations in Daphnia. Molecular responses of biological pathways and gene- Ecology 26, 5003-5015. families in Daphnia exposed to single and Herrmann M, Ravindran SP, Schwenk mixed stressors. Environ Sci Technol 48, K, Cordellier M (2018) Population 3513-3522. transcriptomics in Daphnia: the role of Declerck S, De Meester L (2003) thermal selection. Molecular Ecology 27, Impact of fish predation on coexisting 387-402. Daphnia taxa: a partial test of the temporal Herzog Q, Rabus M, Ribeiro BW, hybrid superiority hypothesis. In: Aquatic Laforsch C (2016) Inducible Defenses with Biodiversity, pp. 83-94. Springer. a" Twist": Daphnia barbata Abandons

20 bioRxiv preprint doi: https://doi.org/10.1101/503904; this version posted June 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

Bilateral Symmetry in Response to an Miner BE, De Meester L, Pfrender Ancient Predator. Plos One 11, e0148556. ME, Lampert W, Hairston NG (2012) Linking Huylmans AK, López Ezquerra A, genes to communities and ecosystems: Parsch J, Cordellier M (2016) De Novo Daphnia as an ecogenomic model. Transcriptome Assembly and Sex-Biased Proceedings of the Royal Society B- Gene Expression in the Cyclical Biological Sciences 279, 1873-1882. Parthenogenetic Daphnia galeata. Mitchell MD, Bairos-Novak KR, Genome Biology and Evolution 8, 3120- Ferrari MC (2017) Mechanisms underlying 3139. the control of responses to predator Klüttgen B, Dülmer U, Engels M, odours in aquatic prey. Journal of Ratte HT (1994) ADaM, an artificial Experimental Biology 220, 1937-1946. freshwater for the culture of zooplankton. Morgans CL, Ord TJ (2013) Natural Water Research 28, 743-746. selection in novel environments: predation Kuchta SR, Svensson EI (2014) selects for background matching in the Predator-mediated natural selection on body colour of a land fish. Animal the wings of the damselfly Calopteryx Behaviour 86, 1241-1249. splendens: differences in selection among Okonechnikov K, Conesa A, García- trait types. The American naturalist 184, Alcalde F (2016) Qualimap 2: advanced 91-109. multi-sample quality control for high- Laforsch C, Tollrian R (2004) Inducible throughput sequencing data. defenses in multipredator environments: Bioinformatics 32, 292-294. cyclomorphosis in Daphnia cucullata. Orsini L, Brown JB, Shams Solari O, et Ecology 85, 2302-2311. al. (2018) Early transcriptional response Lampert W (2011) Daphnia: pathways in Daphnia magna are development of a model organism in coordinated in networks of crustacean‐ ecology and evolution. International specific genes. Molecular Ecology 27, 886- Ecology Institute, Oldendorf/Luhe. 897. Langfelder P, Horvath S (2008) Otte KA, Fröhlich T, Arnold GJ, WGCNA: an R package for weighted Laforsch C (2014) Proteomic analysis of correlation network analysis. BMC Daphnia magna hints at molecular Bioinformatics 9, 559. pathways involved in defensive plastic Li L, Stoeckert CJ, Roos DS (2003) responses. BMC Genomics 15, 306. OrthoMCL: identification of ortholog Otte KA, Schrank I, Fröhlich T, Arnold groups for eukaryotic genomes. Genome GJ, Laforsch C (2015) Interclonal proteomic Research 13, 2178-2189. responses to predator exposure in Daphnia Love MI, Huber W, Anders S (2014) magna may depend on predator Moderated estimation of fold change and composition of habitats. Molecular Ecology dispersion for RNA-seq data with DESeq2. 24, 3901-3917. Genome Biology 15, 550. Pearson WR (2013) An introduction Macháček J (1991) Indirect effect of to sequence similarity (“homology”) planktivorous fish on the growth and searching. Current Protocols in reproduction of Daphnia galeata. In: Bioinformatics 42, 3.1. 1-3.1. 8. Biology of Cladocera, pp. 193-197. Picado A, Chankova S, Fernandes A, Springer. et al. (2007) Genetic variability in Daphnia magna and ecotoxicological evaluation.

21 bioRxiv preprint doi: https://doi.org/10.1101/503904; this version posted June 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

Ecotoxicology and Environmental Safety cyanobacterial protease inhibitors. BMC 67, 406-410. Physiology 10, 6. Rabus M, Söllradl T, Clausen- Sedlazeck FJ, Rescheneder P, Von Schaumann H, Laforsch C (2013) Haeseler A (2013) NextGenMap: fast and Uncovering ultrastructural defences in accurate read mapping in highly Daphnia magna–an interdisciplinary polymorphic genomes. Bioinformatics 29, approach to assess the predator-induced 2790-2791. fortification of the carapace. Plos One 8, Stibor H (1992) Predator induced life- e67856. history shifts in a freshwater cladoceran. Ravindran SP, Herrmann M, Oecologia 92, 162-165. Cordellier M (2019) Contrasting patterns of Stibor H, Lüning J (1994) Predator- divergence at the regulatory and sequence Induced Phenotypic Variation in the level in European Daphnia galeata natural Pattern of Growth and Reproduction in populations. Ecology and Evolution. Daphnia hyalina (Crustacea: Cladocera). Riessen HP (1999) Predator-induced Functional Ecology 8, 97-101. life history shifts in Daphnia: a synthesis of Stich HB, Lampert W (1981) Predator studies using meta-analysis. Canadian Evasion as an Explanation of Diurnal Journal of Fisheries and Aquatic Sciences Vertical Migration by Zooplankton. Nature 56, 2487-2494. 293, 396-398. Rozenberg A, Parida M, Leese F, et al. Subramanian A, Tamayo P, Mootha (2015) Transcriptional profiling of VK, et al. (2005) Gene set enrichment predator-induced phenotypic plasticity in analysis: A knowledge-based approach for Daphnia pulex. Frontiers in Zoology 12. interpreting genome-wide expression Schoeppner NM, Relyea RA (2009) profiles. Proceedings of the National Interpreting the smells of predation: how Academy of Sciences of the United States of alarm cues and kairomones induce America 102, 15545-15550. different prey defences. Functional Ecology Tams V, Lüneburg J, Seddar L, 23, 1114-1121. Detampel J-P, Cordellier M (2018) Schwarzenberger A, Courts C, von Intraspecific phenotypic variation in life Elert E (2009) Target gene approaches: history traits of Daphnia galeata Gene expression in Daphnia magna populations in response to fish exposed to predator-borne kairomones or kairomones. Peerj 6, e5746. to microcystin-producing and microcystin- Tarrant A, Baumgartner M, Hansen B, free Microcystis aeruginosa. BMC et al. (2014) Transcriptional profiling of Genomics 10, 527. reproductive development, lipid storage Schwarzenberger A, Küster CJ, Von and molting throughout the last juvenile Elert E (2012) Molecular mechanisms of stage of the marine copepod Calanus tolerance to cyanobacterial protease finmarchicus. . Frontiers in Zoology 11, 91. inhibitors revealed by clonal differences in Team RC (2017) R: A Language and Daphnia magna. Molecular Ecology 21, Environment for Statistical Computing. R 4898-4911. Foundation for Statistical Computing, Schwarzenberger A, Zitt A, Kroth P, Vienna, Austria. Mueller S, Von Elert E (2010) Gene Timmermans M, Roelofs D, Nota B, expression and activity of digestive Ylstra B, Holmstrup M (2009) Sugar sweet proteases in Daphnia: effects of springtails: on the transcriptional response

22 bioRxiv preprint doi: https://doi.org/10.1101/503904; this version posted June 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

of Folsomia candida (Collembola) to response to chemical cues from predators. desiccation stress. Insect Molecular Biology Oikos, 385-392. 18, 737–746. Wickham H (2009) ggplot2: Elegant Tollrian R (1995) Predator induced Graphics for Data Analysis Springer, New morphological defenses: costs, life history York. shifts, and maternal effects in Daphnia Wolinska J, Löffler A, Spaak P (2007) pulex. Ecology 76, 1691-1705. Taxon-specific reaction norms to predator Tollrian R, Harvell CD (1999) The cues in a hybrid Daphnia complex. ecology and evolution of inducible defenses Freshwater Biology 52, 1198-1209. Princeton University Press. Yin MB, Laforsch C, Lohr JN, Wolinska Van Straalen N (2003.) Ecotoxicology J (2011) Predator-induced defense makes Becomes Stress Ecology. Environ Sci Daphnia more vulnerable to parasites. Technol 37, 324A-330A. Evolution 65, 1482-1488. Weider LJ, Pijanowska J (1993) Plasticity of Daphnia life histories in

23 bioRxiv preprint doi: https://doi.org/10.1101/503904; this version posted June 29, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

Appendix Appendix 1. Principal component (PC) plot of the biological RNA-seq samples of D. galeata. (A) genotype M6 (B) genotype M9. ‘Orange’: control environment. ‘Blue’: fish environment (predation risk). Appendix 2. List of all differentially expressed transcripts (DETs) in D. galeata in response to fish kairomone (predation risk). The list contains 125 predation risk-related DETs including co- expression module, OrthoMCL cluster information and GO term annotation. Hub-genes are highlighted in bold. Appendix 3. Heatmap of correlation values of module eigengenes and external traits genotype and predation risk. Red and blue indicate a positive and negative correlation of the module with the respective trait. Darker hues indicate higher correlation values. Appendix 4. List of significantly enriched gene ontology (GO) terms for the predation risk- related dataset of D. galeata. Only significantly enriched GO terms are shown in ascending order (classicFisher <0.05). GO IDs which occur more than once are highlighted in grey. Appendix 5. List of significantly enriched gene ontology (GO) terms for the genotype-related dataset of D. galeata. Only significantly enriched GO terms are shown in ascending order (classicFisher <0.05). GO IDs which occur more than once are highlighted in grey.